The purpose of this paper is to show the potential of self‐organized semantic storage services. The semantic web has provided a vision of how to build the applications of the future. A software component dedicated to the storage and retrieval of semantic information is an important but generic part of these applications. Apart from mere functionality, these storage components also have to provide good performance regarding the non‐functional requirements scalability, adaptability and robustness. Distributing the task of storing and querying semantic information onto multiple computers is a way of achieving this performance. However, the distribution of a task onto a set of computers connected using a communication network is not trivial. One solution is self‐organized technologies, where no central entity coordinates the system's operation.
Based on the available literature on large‐scale semantic storage systems, the paper analyzes the underlying distribution algorithm, with special focus on the properties of semantic information and corresponding queries. The paper compares the approaches and identify their shortcomings.
All analyzed approaches and their underlying technologies were unable to distribute large amounts of semantic information and queries in a generic way while still being able to react on changing network infrastructure. Nonetheless, as each concept represented a unique trade‐off between these goals, the paper points out how self‐organization is crucial to perform well at least in a subset of them.
The contribution of this paper is a literature review aimed at showing the potential of self‐organized semantic storage services. A case is made for self‐organization in a distributed storage system as the key to excellence in the relevant non‐functional requirements: scalability, adaptability and robustness.
